2006-11-15 Roman Kennke <kennke@aicas.com>

	* gnu/javax/swing/text/html/parser/support/Parser.java
	(_handleText): Consume whitespace directly before a closing tag.
	(restOfTag): Consume whitespace directly after opening.
	* gnu/javax/swing/text/html/parser/support/textPreProcessor.java
	(preprocess): Don't perform array boundary checking by
	catch AIOOBE, instead check the boundary in loop condition.
	* gnu/javax/swing/text/html/parser/support/low/Constants.java
	(TAG_CLOSE): New constants. Describes the token pattern for
	a closing tag.

3 files changed, 27 insertions, 16 deletions

diff --git a/gnu/javax/swing/text/html/parser/support/Parser.java b/gnu/javax/swing/text/html/parser/support/Parser.java
index f1f25fad0..9c10f68fc 100644
--- a/gnu/javax/swing/text/html/parser/support/Parser.java
+++ b/gnu/javax/swing/text/html/parser/support/Parser.java
@@ -659,7 +659,10 @@ public class Parser
     else
       text = textProcessor.preprocess(buffer);
 
-    if (text != null && text.length > 0)
+    if (text != null && text.length > 0
+        // According to the specs we need to discard whitespace immediately
+        // before a closing tag.
+        && (text.length > 1 || (text[0] == ' ' && ! TAG_CLOSE.matches(this))))
       {
         TagElement pcdata = new TagElement(dtd.getElement("#pcdata"));
         if ((text.length > 1 && text[0] != ' ')
@@ -1460,7 +1463,12 @@ public class Parser
         if (te.getElement().type == DTDConstants.EMPTY)
           _handleEmptyTag(te);
         else
-          _handleStartTag(te);
+          {
+            // According to the specs we need to consume whitespace following
+            // immediately after a opening tag.
+            optional(WS);
+            _handleStartTag(te);
+          }
       }
   }
 
diff --git a/gnu/javax/swing/text/html/parser/support/low/Constants.java b/gnu/javax/swing/text/html/parser/support/low/Constants.java
index 283d32385..5416582ad 100644
--- a/gnu/javax/swing/text/html/parser/support/low/Constants.java
+++ b/gnu/javax/swing/text/html/parser/support/low/Constants.java
@@ -209,6 +209,17 @@ public class Constants
                 }
                );
 
+  /**
+   * Ordinary HTML tag closing pattern.
+   */
+  public static final pattern TAG_CLOSE =
+    new pattern(new node[]
+                {
+                  new node(BEGIN), new node(WS, true), new node(SLASH),
+                  new node(WS, true), new node(NUMTOKEN)
+                }
+               );
+
   /* Special tokens */
 
   /**
diff --git a/gnu/javax/swing/text/html/parser/support/textPreProcessor.java b/gnu/javax/swing/text/html/parser/support/textPreProcessor.java
index b81275b1f..6fd79e258 100644
--- a/gnu/javax/swing/text/html/parser/support/textPreProcessor.java
+++ b/gnu/javax/swing/text/html/parser/support/textPreProcessor.java
@@ -65,22 +65,14 @@ public class textPreProcessor
     int b = text.length - 1;
 
     // Remove leading/trailing whitespace, leaving at most one character
-    try
-      {
-        while (Constants.bWHITESPACE.get(text[a])
-               && Constants.bWHITESPACE.get(text[a + 1]))
-          a++;
+    int len = text.length;
+    while (a + 1 < len && Constants.bWHITESPACE.get(text[a])
+           && Constants.bWHITESPACE.get(text[a + 1]))
+      a++;
 
-        while (b > a && Constants.bWHITESPACE.get(text[b])
+    while (b > a && Constants.bWHITESPACE.get(text[b])
                && Constants.bWHITESPACE.get(text[b - 1]))
-          b--;
-      }
-    catch (ArrayIndexOutOfBoundsException sx)
-      {
-        // A text fragment, consisting from spaces and line breaks only,
-        // mutates into single space.
-        return new char[] { ' ' };
-      }
+      b--;
 
     a_text.setLength(0);